Rcwl_GATK4

Introduction

The GATK4 best practices for germline short variant discovery was implemented with Workflow Description Language (WDL), which is similar to cwl and requires cromwell to run the pipelines.

Here is the details of the pipeline: https://software.broadinstitute.org/gatk/best-practices/workflow?id=11145

Germline short variant discovery (SNPs + Indels)

The germline pipelines include 4 steps in WDL, paired fastq to ubam, GATK alignment, variant calling by HaplotypeCaller and joint genotyping. We wrapped the GATK pipelines to 3 steps using Rcwl for different numbers of computing nodes requirements. The wrapped pipelines can help to assign inputs to the input JSON templates and glob results from the cromwell outputs.

First, we need to load the packages and pipelines.

library(Rcwl)
library(RcwlPipelines)
data("GAlign")
data("hapCall")
data("jdCall")

GATK Alignment

We wrapped the steps from raw fastqs to analysis-ready BAM file into GAlign pipeline. Here is the short summary of the pipeline.

short(GAlign)

inputs:
- fastq1
- fastq2
- readGroup
- sampleName
- library
- platunit
- platform
- center
- tmpl1
- wdl1
- tmpl2
- wdl2
- cromwell
outputs:
- bamlog
- outdir
steps:
- fqJson
- fq2ubam
- ubam2bamJson
- align
- mvOut

For the inputList, we need to assign the fastqs files and read groups for each sample. The inputs can be multiple items separated by comma if there are more than one read groups for each sample. The input templates and WDL scripts can be assigned in the paramList and the reference and other GATK bundle files should be changed to your local version of files accordingly in the local json files. The path to the cromwell binary file is also required. Here is an example.

tmpl1 <- system.file(package="RcwlPipelines", "GATK4/seq-format-conversion/paired-fastq-to-unmapped-bam.inputs.json")
tmpl2 <- system.file(package="RcwlPipelines", "GATK4/gatk4-data-processing/processing-for-variant-discovery-gatk4.hg38.wgs.inputs.local.json")
wdl1 <- system.file(package="RcwlPipelines", "GATK4/seq-format-conversion/paired-fastq-to-unmapped-bam.wdl")
wdl2 <- system.file(package="RcwlPipelines", "GATK4/gatk4-data-processing/processing-for-variant-discovery-gatk4.wdl")

inputList <- list(fastq1=list(normal="/home/qhu/workspace/projects/RPipe/data/DNASeq/data/normal_1.fq",
                              tumor="/home/qhu/workspace/projects/RPipe/data/DNASeq/data/tumor_1.fq"),
                  fastq2=list(normal="/home/qhu/workspace/projects/RPipe/data/DNASeq/data/normal_2.fq",
                              tumor="/home/qhu/workspace/projects/RPipe/data/DNASeq/data/tumor_2.fq"),
                  readGroup=list("normal.1", "tumor.1"),
                  sampleName=list("normal", "tumor"),
                  library=list("normal", "tumor"),
                  platunit=list("normal", "tumor"),
                  platform=list("illumina", "illumina"),
                  center=list("rpccc", "rpccc"))
paramList <- list(tmpl1=tmpl1,
                  wdl1=wdl1,
                  tmpl2=tmpl2,
                  wdl2=wdl2,
                  cromwell="/mnt/lustre/users/qhu/software/cromwell-36.jar")
r1 <- runCWLBatch(GAlign, wdir="output/BAM", inputList, paramList,
                  BatchtoolsParam(workers = 2, cluster="sge",
                                  template = "/rpcc/bioinformatics/sge_centos7.tmpl",
                                  resources = list(threads = 16,
                                                   queue = "centos7.q")),
                  stderr="")

The outputs were globbed from the cromwell execution folder.

list.files("output/BAM/normal", recursive = TRUE)

[1] "output/normal.hg38.bai"                        
[2] "output/normal.hg38.bam"                        
[3] "output/normal.hg38.bam.md5"                    
[4] "output/normal.hg38.duplicate_metrics"          
[5] "output/normal.hg38.recal_data.csv"             
[6] "processing-for-variant-discovery-gatk4.wdl.log"

HaplotypeCaller

This step takes the BAM files as input and each BAM file will be assigned to different computing node. Also the json template file need to be modified to the correct GATK bundle paths at first.

wdl3 <- system.file(package="RcwlPipelines", "GATK4/gatk4-germline-snps-indels/haplotypecaller-gvcf-gatk4.wdl")
tmpl3 <- system.file(package="RcwlPipelines", "GATK4/gatk4-germline-snps-indels/haplotypecaller-gvcf-gatk4.hg38.inputs.local.json")

bams <- list(normal = normalizePath("output/BAM/normal/output/normal.hg38.bam"),
             tumor = normalizePath("output/BAM/tumor/output/tumor.hg38.bam"))
inputList <- list(bam = bams)
paramList <- list(intervals = normalizePath("output/interval.txt"),
                  cromwell = "/mnt/lustre/users/qhu/software/cromwell-36.jar",
                  wdl = wdl3,
                  tmpl = tmpl3)

r2 <- runCWLBatch(hapCall, wdir="output/GATK", inputList, paramList,
                  BatchtoolsParam(workers = 2, cluster="sge",
                                  template = "/rpcc/bioinformatics/sge_centos7.tmpl",
                                  resources = list(threads = 16,
                                                   queue = "centos7.q")),
                  stderr="")

Here are the outputs:

list.files("output/GATK/normal", recursive = TRUE)

[1] "haplotypecaller-gvcf-gatk4.wdl.log"
[2] "output/normal.hg38.g.vcf.gz"       
[3] "output/normal.hg38.g.vcf.gz.tbi"   

Joint Discovery

The joint genotyping step will combine the gvcf files and then call variants in all samples, so only one computing node is required. Multiple values or files of the samples need to be pasted by comma for each input in the inputList. The paths of the local bundle files are also need to be added to the json template file.

wdl4 <- system.file(package="RcwlPipelines", "GATK4/gatk4-germline-snps-indels/joint-discovery-gatk4-local.wdl")
tmpl4 <- system.file(package="RcwlPipelines", "GATK4/gatk4-germline-snps-indels/joint-discovery-gatk4-local.hg38.wgs.inputs.json")

inputList <- list(sampleName = list(test="normal,tumor"),
                  gvcf = list(test="/home/qhu/workspace/projects/RPipe/data/DNASeq/output/GATK/normal/output/normal.hg38.g.vcf.gz,/home/qhu/workspace/projects/RPipe/data/DNASeq/output/GATK/tumor/output/tumor.hg38.g.vcf.gz"))
## inputList <- list(sampleName = list(test="NA12878"),
##                   gvcf = list(test="/home/qhu/workspace/projects/RPipe/data/DNASeq/output/GATK/NA12878.g.vcf.gz"))

paramList <- list(callsetName = "test",
                  intervals = "/home/qhu/workspace/projects/RPipe/data/DNASeq/output/interval.21.interval_list",
                  unpadded_intervals = "/home/qhu/workspace/projects/RPipe/data/DNASeq/output/interval.21.intervals",
                  tmpl = tmpl4,
                  cromwell = "/mnt/lustre/users/qhu/software/cromwell-36.jar",
                  wdl = wdl4)

r3 <- runCWLBatch(jdCall, wdir="output/GATK", inputList, paramList,
                  BatchtoolsParam(workers = 1, cluster="sge",
                                  template = "/rpcc/bioinformatics/sge_centos7.tmpl",
                                  resources = list(threads = 16,
                                                   queue = "centos7.q")),
                  stderr="")

Here are the final outputs:

list.files("output/GATK/test", recursive = TRUE)

[1] "joint-discovery-gatk4-local.wdl.log"        
[2] "output/out.intervals"                       
[3] "output/test.variant_calling_detail_metrics" 
[4] "output/test.variant_calling_summary_metrics"
[5] "output/test.vcf.gz"                         
[6] "output/test.vcf.gz.tbi"                     

Session information

sessionInfo()

R version 3.5.2 Patched (2018-12-31 r75935)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS release 6.4 (Final)

Matrix products: default
BLAS: /home/qhu/usr/R-3.5/lib64/R/lib/libRblas.so
LAPACK: /home/qhu/usr/R-3.5/lib64/R/lib/libRlapack.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] parallel  stats4    stats     graphics  grDevices utils     datasets 
[8] methods   base     

other attached packages:
[1] rmarkdown_1.11           RcwlPipelines_0.0.0.9000
[3] jsonlite_1.6             BiocParallel_1.16.2     
[5] Rcwl_0.99.7              S4Vectors_0.20.1        
[7] BiocGenerics_0.28.0      yaml_2.2.0              

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.0         tidyr_0.8.2        prettyunits_1.0.2 
 [4] visNetwork_2.0.5   assertthat_0.2.0   rprojroot_1.3-2   
 [7] digest_0.6.18      R6_2.3.0           plyr_1.8.4        
[10] backports_1.1.3    evaluate_0.12      highr_0.7         
[13] ggplot2_3.1.0      pillar_1.3.1       rlang_0.3.0.1     
[16] progress_1.2.0     lazyeval_0.2.1     rstudioapi_0.8    
[19] data.table_1.11.8  whisker_0.3-2      R.oo_1.22.0       
[22] R.utils_2.7.0      checkmate_1.8.5    DiagrammeR_1.0.0  
[25] downloader_0.4     readr_1.3.1        stringr_1.3.1     
[28] htmlwidgets_1.3    igraph_1.2.2       munsell_0.5.0     
[31] compiler_3.5.2     influenceR_0.1.0   rgexf_0.15.3      
[34] xfun_0.4           pkgconfig_2.0.2    htmltools_0.3.6   
[37] tidyselect_0.2.5   tibble_1.4.2       gridExtra_2.3     
[40] workflowr_1.1.1    batchtools_0.9.11  XML_3.98-1.16     
[43] viridisLite_0.3.0  crayon_1.3.4       dplyr_0.7.8       
[46] withr_2.1.2        R.methodsS3_1.7.1  rappdirs_0.3.1    
[49] grid_3.5.2         gtable_0.2.0       git2r_0.23.0      
[52] magrittr_1.5       scales_1.0.0       stringi_1.2.4     
[55] debugme_1.1.0      viridis_0.5.1      bindrcpp_0.2.2    
[58] brew_1.0-6         RColorBrewer_1.1-2 tools_3.5.2       
[61] glue_1.3.0         purrr_0.2.5        hms_0.4.2         
[64] Rook_1.1-1         colorspace_1.3-2   base64url_1.4     
[67] knitr_1.21         bindr_0.1.1